Convert the source to TypeScript

[mapitman]

Convert the source under lib/ and the gettyimages-api.js entry point from JavaScript to TypeScript.

Why

The main argument is that this collapses work already planned in this milestone, rather than adding to it.

It provides the build step #64 needs. Dual ESM and CommonJS output requires a build. tsc is that build. Without this conversion, #64 adds a build step purely to transform module syntax, which is a lot of machinery for very little gain.

It removes the hand-written declarations. gettyimages-api.d.ts is 572 lines describing roughly 2,000 lines of untyped JavaScript, and nothing enforces that the two agree. types-test.ts checks that the declarations compile; it does not check them against the implementation. Add a builder method and forget the declaration file, and no test fails. Generated declarations cannot drift.

It verifies #65. Marking execute() as async is a contract change across 14 methods. The compiler confirms it landed consistently.

It deletes the WeakMap boilerplate. Every class carries roughly 20 lines of this pattern:

const _ids = new WeakMap();
set ids(value) { _ids.set(this, value); }
get ids() { return _ids.get(this); }

TypeScript's private fields replace it. lib/images.js is 80 lines today and should land near 45.

Scope

• 18 files under lib/, plus gettyimages-api.js. Roughly 2,000 lines, of which about half is repetitive fluent builders that convert mechanically.
• Roughly 20 test files, 1,800 lines.
• Replace jsconfig.json with a real tsconfig.json, and generate declarations with declaration: true.
• Delete the hand-written gettyimages-api.d.ts once the generated output matches it. Keep types-test.ts as a consumer-facing check against the generated declarations.
• Replace the WeakMap private-field pattern with private fields.
• lib/webhelper.js uses the WebHelper function-constructor style, unlike the class style used everywhere else. Normalize it during the conversion.

Out of scope

Typing the API responses. Every execute() returns Promise<any> today, and it should continue to after this change. Modeling the Getty API response shapes is a much larger job than converting the request builders, and it belongs in its own issue.

Sequencing

Do this first in the milestone, ahead of #65 and #64.

• Expose execute() as async #65 (async execute()) is easier afterwards, because the compiler verifies the change across all 14 methods.
• Support both ESM and CommonJS #64 (dual ESM and CommonJS) becomes mostly a tsconfig output setting plus the exports map, because the build step already exists.

Doing #64 or #65 in JavaScript first means redoing parts of both.

Notes

• This is not user-visible. Consumers already have typings, as of the declarations added for Typescript Support? #56. This change buys maintainer confidence and enables the rest of the milestone; it does not add consumer capability on its own.
• Cost to weigh: contributors will need TypeScript to submit a PR. That is a real, if modest, barrier for an SDK that takes occasional outside contributions.